Dynamic Disease Management in Trachymyrmex Fungus-Growing Ants (Attini: Formicidae)

The University of Chicago

Dynamic Disease Management in Trachymyrmex Fungus-Growing Ants (Attini: Formicidae). Author(s): Hermógenes Fernández-Marín, Gaspar Bruner, Ernesto B. Gomez, David R. Nash, Jacobus J. Boomsma, and William T. Wcislo Source: The American Naturalist, Vol. 181, No. 4 (April 2013), pp. 571-582 Published by: The University of Chicago Press for The American Society of Naturalists Stable URL: http://www.jstor.org/stable/10.1086/669664 . Accessed: 31/07/2015 11:05

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

The University of Chicago Press, The American Society of Naturalists, The University of Chicago are collaborating with JSTOR to digitize, preserve and extend access to The American Naturalist.

http://www.jstor.org

This content downloaded from 160.111.134.19 on Fri, 31 Jul 2015 11:05:40 AM All use subject to JSTOR Terms and Conditions vol. 181, no. 4 the american naturalist april 2013

Natural History Note Dynamic Disease Management in Trachymyrmex Fungus-Growing Ants (Attini: Formicidae)

Hermo´genes Ferna´ndez-Marı´n,1,2,* Gaspar Bruner,1 Ernesto B. Gomez,1 David R. Nash,2 Jacobus J. Boomsma,2 and William T. Wcislo1,†

1. Smithsonian Tropical Research Institute, Apartado Postal 0843-03092, Balboa, Anco´n, Panama, or Box Unit 9100, Box 0948, DPO AA 34002-0998; 2. Centre for Social Evolution, Department of Biology, University of Copenhagen, Universitetsparken 15, 2100 Copenhagen, Denmark Submitted February 29, 2012; Accepted September 18, 2012; Electronically published February 27, 2013 Dryad data: http://dx.doi.org/10.5061/dryad.c2090.

vide interesting models to explore the evolutionary trade- abstract: Multipartner mutualisms have potentially complex dy- offs that may lead to one defense being prioritized over namics, with compensatory responses when one partner is lost or relegated to a minor role. Fungus-growing ants (Attini) are mutualistic another, because escape from enemies by fleeing is gen- associates of basidiomycete fungi and antibiotic-producing actino- erally not possible. Plants, for example, deploy a diverse mycete bacteria; the former are attacked by specialized fungi (Esco- array of biochemical, morphological, and biotic defenses vopsis) and diverse generalist microbes. Ants deploy biochemical de- derived from symbionts, even among closely related spe- fenses from bacteria and metapleural glands (MGs) and express cies (reviewed in Coley and Barone 1996; Stamp 2003; different behaviors to control contaminants. We studied four Trachy- Fine et al. 2006; Agrawal 2007). Each of these alternatives myrmex species that differed in relative abundance of actinomycetes carries a cost (Heil 2002), and thus many species confront to understand interactions among antimicrobial tactics that are con- tingent on the nature of infection. MG grooming rate and actinomycete energetic trade-offs associated with investment in one de- abundance were negatively correlated. The two species with high MG fensive mechanism or another or with differential invest- grooming rates or abundant actinomycetes made relatively little use of ment in somatic defenses versus reproductive effort (e.g., behavioral defenses. Conversely, the two species with relatively modest Itino and Itioka 2001; Fincher et al. 2008). Similar trade- biochemical defenses relied heavily on behavior. Trade-offs suggest that offs are known in sponges (e.g., Leong and Pawlik 2011) related species can evolutionarily diverge to rely on different defense and other animals. A comparison of tree and house spar- mechanisms against the same threat. Neither bacterial symbionts nor rows (Passer), for example, showed that tree sparrows ex- MG secretions thus appear to be essential for mounting defenses against the specialized pathogen Escovopsis, but reduced investment in one of hibit a relatively weak nonspecific inflammatory response these defense modes tends to increase investment in the other. to malarial blood parasites and a relatively strong inducible immune response, relative to house sparrows, and vice Keywords: metapleural glands, fungal symbiont, hygiene, grooming, versa (Lee et al. 2006). weeding, Actinomycetales, Attini, Escovopsis. Studies of eusocial insects are especially relevant to understanding trade-offs in antimicrobial strategies. Colony Introduction sizes range from a few related individuals to tens of thousands or even millions, living in crowded conditions The hygienic and therapeutic strategies that a host can use favorable for pathogen transmission (Schmid-Hempel to prevent or combat disease depend on the nature of 1998; Boomsma et al. 2005; Cremer and Sixt 2009). Con- infections and on the availability and costs of alternative siderable evidence now suggests that polyandry may have defense mechanisms. Sessile organisms (e.g., plants) or secondarily evolved in response to disease pressure in eu- temporarily sessile organisms (e.g., nesting animals) pro- social insects with large, long-lived colonies (Hughes and Boomsma 2004; Tarpy and Seeley 2006). In a colony with * Corresponding author. Present address: Instituto de Investigaciones Cientı´ficas a single multiply-mated queen, the genetic diversity in the y Servicios de Alta Tecnologıá, Edificio 219, Ciudad del Saber, Clayton, Panama long-lived chimeras of patrilines remains approximately City, Panama; e-mail: [email protected], [email protected]. † Corresponding author; e-mail: [email protected]. constant throughout the colony’s life span, which may Am. Nat. 2013. Vol. 181, pp. 571–582. ᭧ 2013 by The University of Chicago. facilitate the intracolony transmission of pathogens 0003-0147/2013/18104-53690$15.00. All rights reserved. (Hughes and Boomsma 2006; Mattila and Seeley 2007; DOI: 10.1086/669664 Reber et al. 2008). It is therefore no surprise that eusocial

This content downloaded from 160.111.134.19 on Fri, 31 Jul 2015 11:05:40 AM All use subject to JSTOR Terms and Conditions 572 The American Naturalist animals have evolved a diverse assembly of behavioral and per colony with visible actinomycetes and the frequency biochemical mechanisms to control pathogens (Oi and of MG grooming (Fernańdez-Marıń et al. 2009), suggest- Pereira 1993; Cremer et al. 2007; Reber et al. 2008). Many ing that the simultaneous use of multiple antimicrobial prior studies have focused on one defense strategy over defenses may be too expensive to be a prudent strategy. another, and hence the coordination and integration of That study contrasted representatives of the higher attine these multiple defense strategies is not yet well understood genera Trachymyrmex and Acromyrmex (with visibly abun- (e.g., Boomsma et al. 2005). dant actinomycete bacterial cover in workers) with those The obligate mutualism between attine ants and their of Atta and Sericomyrmex (which lack or have few visible fungal cultivars is ideally suited to explore how ants coor- bacteria; Mueller et al. 2008; Fernańdez-Marıń et al. 2009). dinate and integrate different elements of disease control These comparisons showed that visible bacterial cover was (Weber 1972; Currie et al. 1999b; Bot et al. 2002; Hart et greatly reduced or lost in genera with relatively larger col- al. 2002; Mueller 2002; Little and Currie 2008) and how ony sizes (for a discussion of breakdown of mutualisms, their societies adjust and optimize integrated pest manage- see Sachs and Simms 2006; Fernańdez-Marıń et al. 2009). ment strategies under different or changing social and en- These genera, however, also differ greatly in numerous vironmental conditions (Fernańdez-Marıń et al. 2009). At- other aspects of their natural history, including differences tine ants are members of a mutualistic consortium (sensu in mature-colony size, queen mating frequency, diet, and Kozo-Polyansky [1924] 2010), along with leucocoprinous worker caste differentiation, which made it difficult to gen- or pterulaceous (Basidiomycotina) fungi that are cultivated eralize those prior results. as food, and most species harbor filamentous Actinomy- Here we aim to help resolve this ambiguity with a com- cetales bacteria, which are cultured on their exoskeletons, parative study of four species of Trachymyrmex from cen- to obtain antibiotics active against pathogens (Currie et al. tral Panama that are relatively similar in social structure 1999a, 1999b, 2006; Mueller et al. 2008; Mattoso et al. 2012). and life history but differ in abundance of actinomycetes, The ant fungal gardens are attacked by a parasitic micro- ranging from not visible to highly abundant on the exo- fungus (Escovopsis, Ascomycotina; Currie et al. 2003). skeleton. Specifically, we tested whether (1) the visible Strains of Escovopsis differ in their virulence, and here we abundance of cuticular bacteria across species was in- assess whether ants adjust their antimicrobial tactics ac- versely associated with the frequency of MG grooming, cording to the virulence of the pathogen. Both the cultivar (2) weeding and fungus-planting behaviors might com- and the ants are also attacked by an array of generalist pensate for a reduction in visible abundance of actino- pathogens (see Fernańdez-Marıń et al. 2006, 2009 and ref- mycetes, (3) ant hygienic behavior changes when colonies erences therein). A black yeast occurs in gardens of some are infected with different Escovopsis strains, (4) the size attines that may antagonize the mutualistic actinomycete of the MG reservoir (bulla) increases with a decrease in bacteria and thus reduce the ants’ abilities to suppress Es- visibly abundant actinomycetes, and (5) reduced abun- covopsis infections (Little and Currie 2007, 2008). dance of cuticular actinomycetes is more likely to occur Attine ants actively monitor their fungal cultivar, brood, in species with larger colonies. and nestmates for signs of disease, and they weed the former and groom the latter to remove contaminants (e.g., Currie and Stuart 2001). They also plant pieces of healthy Material and Methods fungal cultivar over contaminated areas of the garden (see Ant Colony Collection and a Prećis of “Definitions of Behaviors” below), which may modify local Trachymyrmex Natural History competitive dynamics. Finally, when they encounter non- cultivar fungal conidia, they apply antimicrobial secretions The four species of Trachymyrmex in this study are abundant from exocrine glands, especially the paired metapleural in central Panama, although studies of their natural history glands (MGs; Fernańdez-Marıń et al. 2006, 2009). MG have been fragmentary (Dijkstra and Boomsma 2003; Fer- secretions and bacterial metabolites have been viewed as nańdez-Marıń et al. 2004, 2006; Hughes et al. 2008; Pe´rez- dual complementary antimicrobial strategies of attine ants Orte´ga et al. 2010). Colonies of Trachymyrmex cornetzi, Tra- (Mueller et al. 2005, 2008; Currie et al. 2006; Fernańdez- chymyrmex sp. 3, and Trachymyrmex zeteki were collected Marıń et al. 2006). Recent work has suggested that some in open and disturbed areas or forested regions, near Gam- members of the antibiotic-producing bacterial community boa and Parque Nacional Soberania in central Panama, that are specific against Escovopsis while others have broad- receive about 1,800 mm rain per year. Colonies of Trachy- spectrum activity (Haeder et al. 2009; Oh et al. 2009; Barke myrmex sp. 10 were collected near Cocoli, in a drier forest et al. 2010; Schoenian et al. 2011; Mattoso et al. 2012). and adjacent open areas that receive about 1,500 mm of Comparative analysis has further revealed a negative in- rain per year (for details of the site, see Condit et al. 2004). terspecific correlation between the proportion of workers Cocoli is near the Pacific coast of Panama City, about 20

This content downloaded from 160.111.134.19 on Fri, 31 Jul 2015 11:05:40 AM All use subject to JSTOR Terms and Conditions Managing Disease in Fungus-Growing Ants 573 km from the Gamboa/Soberania site. At the Cocoli site, that facilitates the suspension of spores and cells in water; there are other attines, including species representing My- Sigma, St. Louis). Dilutions were plated on chitin agar (10 cocepurus, Apterostigma, Myrmicocrypta, Cyphomyrmex, Ser- g chitin from crab shells, 0.5 g peptone, 0.1 g yeast extract, 7 icomyrmex, Trachymyrmex, Atta, and Acromyrmex. At this 0.01 g FePO4 H2O, 250 mL filtered seawater, 15 g agar), site, we have no data on cultivar sharing. with 0.2 g/L of Nystatin (Bristol-Myers Squibb) as an an- In all species, the fungus gardens were usually suspended timycotic. Actinomycete colonies were grown at 25ЊC, and from rootlets that entered the chamber, or were attached the number of CFUs was counted 4 weeks later. to the ceiling (see Fernańdez-Marıń et al. 2007), but parts We recorded the prevalence and intensity of Escovopsis of larger gardens contacted the chamber floor. Nests had infections in nests of each ant species by isolating 50 pieces a single entrance, which in mature colonies generally con- of fungus garden per nest (3–5 mm diameter) and culturing nected to a single fungus-garden chamber in T. cornetzi 10 inoculations per petri dish; each petri dish contained and Trachymyrmex sp. 3, one to four chambers in T. zeteki, potato dextrose agar (PDA, 19.5 g/500 mL of sterilized wa- and three to nine chambers in Trachymyrmex sp. 10. To ter) without antibiotics. We counted the number of healthy obtain colony size estimates (numbers of worker), we col- and Escovopsis-infected pieces after 10 days. Escovopsis in- lected 18 nests of Trachymyrmex sp. 10, 17 nests of Tra- fection was identified by mycelium growth and confirmed chymyrmex sp. 3, 20 nests of T. cornetzi, and 128 nests of by conidia production. The number of nest samples per T. zeteki. species was 12 for T. cornetzi,13forTrachymyrmex sp. 10, Nests were excavated with a geology pick and forceps, 3 for Trachymyrmex sp. 3, and 12 for T. zeteki. and when necessary, an electric aspirator was used to collect workers. The fungus garden was carefully collected to Definitions of Behaviors minimize any structural disturbance or introduction of contaminants from the soil. Ants were housed in plastic Self- and allo-grooming behaviors are common hygienic boxes using standard methods (Weber 1972). Three times behaviors of insects but were not included in this study. per week, ants were fed with rice, corn, and pieces of fresh “Self-grooming behavior” is not very specific, and it occurs leaves and supplied with water. All nests were used within when an ant rubs one of several legs across various body 3 weeks after collection. regions (e.g., Jander 1976); “allo-grooming behavior” in- For three colonies of T. zeteki and two colonies each of volves similar limb movements, but the targets are adult T. cornetzi, Trachymyrmex sp. 3, and Trachymyrmex sp. 10, or immature nestmates. we recorded the abundance of visible actinomycetes and “Fungal grooming behavior” involves ants “licking” the measured head width, antennal scape length, pronotum fungus garden surface to remove any particles that are width, and MG bulla length and width for 50–127 workers presumably contaminated (Currie and Stuart 2001). per nest. Morphological measurements were made to “Metapleural gland (MG) grooming” serves to transfer the nearest 0.01 mm with a Leica Wild M10 stereo- MG secretions to a source of infection and is characterized microscope. by a highly stereotypical behavior: worker ants extend their legs to raise the body from the substrate and then flex a foreleg along the femorotibial joint to bring the posterior Abundance of Actinomycetes and Escovopsis surface of the metatarsus into contact with the opening We scored the abundance of visible actinomycetes on 16– of one of the MGs. “Fungus-planting behavior” involves 50 workers from each of 8–15 colonies per species in 2007, a worker cutting a piece of healthy fungus garden and and in 2008 workers were sampled from 10–14 colonies transplanting it to an infected area of a garden, which is per species (see appendix for details). To score actino- distinct from behavior associated with the initiation of new mycete cover, we used a scale developed by Poulsen et al. garden growth on freshly collected substrate. “Weeding (2002, 2003a, 2003b; see also Currie et al. 2003; Fernańdez- behavior” occurs when a piece of the garden is removed Marıń et al. 2009), which ranged from 0 (no visible ac- and placed in a garbage dump away from the garden. tinomycetes) to 12 (exoskeleton completely covered). We also isolated actinomycetes and quantified their abundance Pathogen-Switching Experiments in fungus gardens and on ant workers by counting colony- forming units (CFUs), following standard methods (Biani A total of 12 colonies for each species were used in a et al. 2009), from five randomly selected workers and 0.05 pathogen-switching experiment. From each colony, we g of fungus garden from each of 10 nests. Under a laminar- created a single subcolony of 20 workers, 0.5 g of fungus flow fume hood, the method involved placing a sample in garden, and six brood items (three larvae and three pupae). 10 mL of a solution of sterile water with 0.85% sodium Three of the 12 subcolonies were then infected with each chloride, 0.2% peptone, and 0.05% Tween 80 (a surfactant of the Escovopsis strains isolated from nests of the four

This content downloaded from 160.111.134.19 on Fri, 31 Jul 2015 11:05:40 AM All use subject to JSTOR Terms and Conditions 574 The American Naturalist

Trachymyrmex species (i.e., there were three replicates per overgrown the petri dish (range of excluded replicates per ant species–Escovopsis strain combination). species, 0–6; for details, see appendix). While we cannot A semicircular piece of fungus garden (0.5 g, ∼2.5 cm be certain that the Escovopsis sporulating on the plates was in diameter) with brood was placed on a sterile petri dish not already present in the fungus garden, the morphology (100 mm # 15 mm) with sterile forceps. A piece of para- and color of the Escovopsis germinated corresponded to film containing ∼2 # 10 6 dry Escovopsis conidia was then that inoculated in all cases. gently rubbed over this garden fragment, after which sterile forceps were used to add 20 randomly chosen workers. Statistical Analyses and Voucher Specimens After an acclimation period of around 10 min, we recorded the number of ants grooming the garden and the rates of Many of the measured variables were nonnormally dis- MG grooming, fungus-planting behavior, and weeding be- tributed or showed strong heteroskedasticity, in which case havior per worker for six consecutive periods of 10 min, they were analyzed with nonparametric Kruskal-Wallis with the aid of a 70# stereomicroscope. tests. Pairs of species were then compared by Mann-Whit- We examined Escovopsis prevalence in infrabuccal pellets ney U-tests with Bonferroni-corrected significance levels. after the switching experiments described above. Infra- Because of the lack of a robust species-level phylogeny, we buccal pellets consist of trash collected by the ants, in- treated each species as an independent data point and did cluding abnormal fungal fragments, bacterially infected not implement comparative methods to take phylogenetic items, and other materials collected during grooming. data into account. These items are compacted and stored in the infrabuccal The number of CFUs cultured from fungus fragments cavity just behind the mouthparts and subsequently re- and workers and infrabuccal-pellet germination rates were gurgitated (Bailey 1920). When a colony is infected with compared by means of generalized linear models with fungal conidia, the infrabuccal pellet is primarily com- Poisson and binomial error structure, respectively, cor- posed of infectious conidia (Little et al. 2003, 2006; Fer- rected for overdispersion. To examine whether the size of nańdez-Marıń et al. 2006). Using a stereomicroscope and the MG is related to actinomycete abundance both be- a sterile loop, we gently removed each fresh infrabuccal tween and within species, we examined the allometric re- pellet (4–7 h after infection) from ants from all subcolonies lationship between a composite measure of body size (the and inoculated them onto sterile petri dishes with PDA first principal component from an unrotated principal- medium. Pellets are fragile, so we discarded those that component analysis of loge[head width] and loge[prono- # broke during manipulations (never more than four pellets tum width]) and MG size (loge[bulla diameter bulla per subcolony). Each plate was monitored daily for up to width]). Residuals from this analysis were used to test 10 days, and the presence of Escovopsis (defined by spor- whether there was a relationship between MG size relative ulation) in the pellets was recorded. to body size and visible actinomycete abundance, using a nested ANCOVA. The number of Escovopsis conidia sporulating was log transformed to normalize their distribution Inhibition of Escovopsis sp. Sporulation by and compared between fungal and Escovopsis strains with the Fungal Cultivar two-way ANOVA. To determine whether the fungal symbiont reduced Es- Ant vouchers were deposited in the Museo de Inver- covopsis sporulation,1goffungus garden from 32 colonies tebrados, Universidad de Panama´. of each of the four Trachymyrmex species was placed onto separate sterile petri dishes. Each petri dish also had a ring Results of cotton wool moistened with water placed around its rim to maintain humidity. The petri dishes were then di- The intensity of Escovopsis infection differed significantly vided randomly into four groups of eight per ant species, between the different Trachymyrmex species (table 1; Krus- and each group was inoculated with6.0 # 10 4 dry conidia kal-Wallis test:H p 9.35 , df p 3 ,P p .025 ). The species of Escovopsis from gardens of each of the Trachymyrmex also differed significantly in colony size (H p 20.34 , species in a complete factorial design. Ten days after in- df p 3,P ! .001 ; fig. 1A) and the number of colony-form- oculation, the fungus gardens including Escovopsis were ing units cultured from fungus gardens and workers after suspended in 80 mL of a solution of Tween (0.2%); then, 4 weeks (fig. 1B; generalized linear model: between species: 1 mL of this solution was sampled and transferred to a likelihood ratio (LR)x 2 p 13.1 , df p 3 ,P p .004 ; fungus hemocytometer to count the number of sporulated co- vs. workers: LRx 2 p 1.77 , df p 1 ,P p .184 ; interaction: nidia. Counts were made only of replicates in which Es- LRx 2 p 2.67 , df p 3 ,P p .445 ). Visible actinomycete covopsis was the dominant fungus, and we excluded rep- cover differed significantly both between species (H p licates in which Trichoderma, Rhizopus, or other fungi had 38.0, df p 3 ,P ! .001 ; fig. 1C) and among nests within

This content downloaded from 160.111.134.19 on Fri, 31 Jul 2015 11:05:40 AM All use subject to JSTOR Terms and Conditions Managing Disease in Fungus-Growing Ants 575

Table 1: Summary of Escovopsis infection and disease management strategies for the four Trachymyrmex species (SD ע studied (means Trachymyrmex species

T. cornetzi T. sp. 10 T. sp. 3 T. zeteki H3 P 025. 9.35 5.9 ע 17.8 4.2 ע 10.0 3.4 ע 17.1 7.8 ע Escovopsis infection (%) 43.1 001.! 38.0 45. ע 5.94 23. ע 2.95 0. ע 0. 22. ע Visible actinomycete abundance 2.43 001.! 12.9 7. ע 13.3 6. ע 14.7 9. ע 10.2 1.0 ע Number of workers on garden 12.5 001.! 33.4 26. ע 26. 17. ע 32. 79. ע 5.51 08. ע MG grooming events workerϪ1 hϪ1 .09 001.! 21.4 2. ע 1.17 3. ע 1.79 4. ע 1.1 5. ע Weeding events workerϪ1 hϪ1 3.62 001.! 16.9 12. ע 36. 58. ע 1.55 11. ע 32. 73. ע Fungal planting events workerϪ1 hϪ1 3.34 p Note: Differences between species were assessed with Kruskal-Wallace tests with 3 df (H3). MG metapleural gland. species for Trachymyrmex cornetzi (,,H p 265.3 df p 22 nation from infrabuccal pellets by the different Trachy- P ! .001), Trachymyrmex sp. 3 (H p 93.8 , df p 18 , P ! myrmex species was marginally insignificant (fig. 3; LR .001), and Trachymyrmex zeteki (,,H p 407.4 df p 22 x 2 p 7.38, df p 3 ,P p .061 ), but there was a significant P ! .001), but not for Trachymyrmex sp. 10, where only a difference between different Escovopsis strains (LR x 2 p few individuals in one nest had any visible actinomycetes 8.07, df p 3 ,P p .045 ) and a significant interaction be- (H p 0.516 , df p 27 ,P 1 .5 ). There was no significant tween Escovopsis strain and Trachymyrmex species (LR difference in visible actinomycete abundance between x 2 p 20.04, df p 3 ,P p .018 ). There was a trend for workers examined in 2007 and 2008 for any of the species greater germination of Escovopsis strains when confronted 1 (Kruskal-Wallis test: allP .28 ). with the species from which they were harvested (the com- Trachymyrmex species differed in their strategies to com- binations along the leading diagonal in fig. 3), compared bat pathogen exposure. After infection, the four Trachy- with other species, that was marginally insignificant (LR myrmex showed significant differences in numbers of work- x 2 p 3.68, df p 1 ,P p .055 ), although the proportion ers tending the infected gardens per 10-min period (H p of explained variance was small. 12.85, df p 3 ,P p .005 ), so behavior rates per worker were The size of the MG increased with body size for all four used in the other analyses to remove this effect. The four Trachymyrmex species (fig. 4; nested ANCOVA: F p species differed significantly in their per-worker rates of MG 3, 21.2 106.71,P ! .001 ), but the slope of this relationship differed grooming (H p 33.4 , df p 3 ,P ! .001 ), weeding (H p between species (F p 3.11 ,P p .026 ); at the same body 21.4, df p 3 ,P ! .001 ), and fungus planting (H p 16.9 , 3, 21.2 size, species differed significantly in MG size ( p df p 3,P ! .001 ), but these did not differ between different F3, 21.2 ! strains of Escovopsis (MG grooming:H p 0.56 , df p 3 , 17.88,P .001 ). The relationship between MG size and P 1 .5; weeding:H p 2.59 , df p 3 ,P p .459 ; fungus body size within a species was not significantly different p p planting:H p 0.98 , df p 3 ,P 1 .5 ). between different nests (F5, 21.2 1.72 ,P .127 ). There was no overall relationship between use of MG The relationship between relative MG size and visible grooming and median visible actinomycete abundance actinomycete cover varied across the three species with within three of the species, because one of the two mea- visible actinomycete cover (nested ordinal logistic sures was always invariant (fig. 2). For the fourth species, ANCOVA: LRx 2 p 6.58 , df p 2 ,P p .037 ). There was T. cornetzi, there was a negative relationship that was mar- no correlation between residual MG size and residual ac- ginally insignificant (Spearman’sr p Ϫ0.657 ,n p 8 , tinomycete cover (calculated by subtracting the median .05 ! P ! .10). For the four species combined, however, actinomycete cover for each colony from the actinomycete there was a clear negative association between MG groom- cover of each member of that colony) for T. cornetzi ing rates and visible actinomycete abundance (fig. 2); T. (Spearman’sr p Ϫ0.115 ,n p 61 ,P p .377 ) and Tra- zeteki had the highest level of actinomycete coverage and chymyrmex sp. 3 (r p ϩ0.084 ,n p 112 ,P p .375 ), but showed no MG grooming, while Trachymyrmex sp. 10 had there was a significant positive correlation for T. zeteki the highest rate of MG grooming and no visible actino- (r p ϩ0.170 ,n p 247 ,P p .008 ; fig. 5). mycetes. Across all species, the relationship between MG The number of Escovopsis conidia sporulating was sig- grooming rates and visible actinomycete abundance is well nificantly different in fungus gardens derived from the p described by an indifference curve based on a Cobb-Doug- nests of the different Trachymyrmex species (F3, 73 23.9 , las utility function (orthogonal regression:U p 0.607 ; P p .0005) but did not depend on the species of Trachy- 2 p p p p lack of fit:x 4.46 , df 30 ,P .999 ). myrmex from which the Escovopsis was derived (F3, 73 Difference in the inhibition of Escovopsis spore germi- 1.17,P p .325 ), and there was no interaction between

This content downloaded from 160.111.134.19 on Fri, 31 Jul 2015 11:05:40 AM All use subject to JSTOR Terms and Conditions A

This content downloaded from 160.111.134.19 on Fri, 31 Jul 2015 11:05:40 AM All use subject to JSTOR Terms and Conditions Managing Disease in Fungus-Growing Ants 577

logenetically independent, because Trachymyrmex is para- phyletic (Schultz and Brady 2008), with some species being more closely related to Sericomyrmex than to other Tra- chymyrmex. A current phylogenetic hypothesis places Tra- chymyrmex zeteki in the same clade as Sericomyrmex, with Trachymyrmex cornetzi in another clade (Schultz and Brady 2008). While Sericomyrmex virtually lacks visible actinomycetes, T. zeteki has twice as many as T. cornetzi (table 1). It therefore seems unlikely that phylogenetic constraints are important for disease management strategies; rather, our results suggest that gains and losses of actinomycete cover can evolve quickly. Our observations are potentially complicated by two additional considerations. First, one of the main trade-offs involved Trachymyrmex sp. 10 (with low levels of actinomycete cover) versus T. cornetzi, T. zeteki, and Trachy- myrmex sp. 3 (all with higher levels of cover). This comparison is potentially confounded because the former Figure 2: Negative correlation between median metapleural gland species was collected in a habitat drier than that from (MG) grooming rates by attine workers following exposure to Es- which the latter three species were collected. We have no covopsis spores, compared with the median relative abundance of evidence to date of habitat-specific pathogens attacking visible actinomycete bacteria on the workers’ exoskeleton (Spear- Trachymyrmex. After our studies, we encountered a site p Ϫ ! man’sr 0.618 ,P .001 ). Each symbol represents the data from about halfway between Gamboa and Cocoli where all four a single colony. Overlapping symbols are slightly offset along the X- axis. The dashed line is a fitted indifference curve based on the Cobb- species co-occur sympatrically (H.F.-M., personal obser- Douglas utility function (see text for details). vation), which will enable us to test directly for habitat- specific pathogens. For other fungus-growing ants (Atta), p Escovopsis strain and fungus garden strain (F9, 73 0.298 , phylogenetic studies have shown that the evolutionary his- P p .973; fig. 3B). tory of Escovopsis does not track the evolutionary history of the ants (Taerum et al. 2007). Some strains of Escovopsis that attack Atta colombica, Atta cephalotes, and Atta sexdens Discussion group together in phylogenetic analyses. All of these host Our results show that variation in the relative abundance species have lowland tropical distributions, but A. sexdens of visible actinomycetes is associated with differential de- prefers drier habitats (Weber 1972). Thus, at least for one ployment of MG secretions, weeding, and fungus-planting of the specialized parasites, there is no evidence for habitat- behaviors to control garden disease in Trachymyrmex ants. specific pathogens that could drive observed differences. Such observations support previous work on disease man- Our study used two methods to estimate actinomycete agement by attines, which demonstrated a trade-off be- abundance that yielded apparently contrasting results: T. tween MG grooming rates and actinomycete abundance cornetzi and Trachymyrmex sp. 3 had higher values for among genera, in that Atta and Sericomyrmex have coopted actinomycete cover than did Trachymyrmex sp. 10, which MG secretions to target a specialist pathogen (Escovopsis) showed virtually no actinomycete cover, yet workers of all that is controlled by actinomycete antibiotics in other at- three species showed similar numbers of colony-forming tines (Fernańdez-Marıń et al. 2009). units (CFUs). Counting CFUs is a culture-dependent Here we document a species-level example of this trade- method that is well known to underestimate bacterial off between MG secretions and actinobacterial antibiotics. abundance, because the culture medium is likely to sustain We do not know to what extent our comparisons are phy- growth of some bacteria but not others. Instead, a culture-

Figure 1: Measures of worker size, colony size, and actinomycete abundance in Panamanian Trachymyrmex. A, Box plots showing the distribution in the size of Trachymyrmex workers (H p head width, P p pronotum width) and their metapleural glands (MGs; W p MG SE) of actinomycete bacteria developing ע bulla width, L p MG bulla length), along with colony size (C). B, Colony-forming units (mean after 3 weeks of culture of fragments of fungus garden (open bars) or worker ants (ﬁlled bars). C, Variation of visible actinomycete cover within and between Trachymyrmex species. Each box plot shows the distribution of cover classes within a single colony of the appropriate species. For box plots, the narrowest section of each box represents the median value, with the extremes of the box giving the 25% and 75% quartiles and the whiskers giving the extremes of the distribution. Means are marked with diamonds.

This content downloaded from 160.111.134.19 on Fri, 31 Jul 2015 11:05:40 AM All use subject to JSTOR Terms and Conditions 578 The American Naturalist

A B 0.5 8.5

T. cornetzi 8.0

7.5 T. sp. 10

7.0

T. sp. 3 Ant species 6.5

T. zeteki 10 6.0 Log (Number of sporulating conidia) 5.5 sp. 3 sp. 10 T. sp. 3 T. T. zeteki sp. 10 T. T. cornetzi T. T. zeteki Escovopsis source T. cornetzi Ant species

Figure 3: A, Pellet germination rates. Escovopsis strains derived from each Trachymyrmex species were used to infect every other Trachymyrmex ,1SE.Bע species. The diameter of the gray circle represents the mean rate of germination, with the black and white circles representing Box plots showing the number of sporulating conidia produced by Escovopsis derived from the fungus gardens of four Trachymyrmex species (“Escovopsis”) when used to infect garden fragments of their own and the other three Trachymyrmex species (“Ant species”). The narrowest section of each box represents the median value, with the extremes of the box giving the 25% and 75% quartiles and the whiskers giving the extremes of the distribution. Means are marked with diamonds. free method, such as visually estimating the white bacterial ing and fungal-planting behaviors, compared with T. zeteki cover on workers’ exoskeletons or next-generation se- (with abundant actinomycetes and relatively low rates of quencing, will possibly yield a more reliable result (e.g., MG grooming) and Trachymyrmex sp. 10 (nearly no visible Cafaro and Currie 2005; Currie et al. 2003; Kost et al. actinomycetes but high rates of MG grooming), consistent 2007; Mueller et al. 2008; Haeder et al. 2009; Sen et al. with the second hypothesis. While nothing is known of 2009; Barke et al. 2010). the dose-dependent effects of actinomycete antibiotics or When we removed Trachymyrmex sp. 10 from our anal- MG compounds, it is interesting to note that even closely yses, the negative correlation between MG grooming rate related Trachymyrmex species differ in their glandular and actinomycete abundance was not significant, sug- chemistry (Adams et al. 2012), although whether this is gesting that MG secretions may play a minimal role in the case for MG secretions remains to be tested. controlling Escovopsis in some Trachymyrmex. Overall, the Complementarity of multiple defenses in attine ant pub- differential abundance of actinomycetes among Trachy- lic health management raises new questions about the rel- myrmex species might thus be explained by two hypotheses ative metabolic costs of alternative defenses, as suggested that are not mutually exclusive. First, different species may by Fernańdez-Marıń et al. (2009). This complementarity have actinomycetes that produce antibiotics with differing hypothesis was developed for the interaction between MG degrees of efficacy in controlling Escovopsis, and they incur secretions and actinomycete maintenance, but our results different costs to maintain the bacteria, so that species imply that similar arguments may apply to trade-offs with could have relatively few actinomycetes that produce more weeding and grooming behaviors, although we lack data potent antibiotics, but this has not been tested. Second, on the metabolic costs associated with behavior. The func- Trachymyrmex species could complement their actino- tional role of weeding and grooming as an alternative to mycete antibiotics with other hygienic behaviors or sources employing antimicrobial compounds is clear, because this of antimicrobial compounds. We found that T. cornetzi behavior removes pathogens and infected fungal-garden and Trachymyrmex sp. 3, both species with intermediate pieces (Bailey 1920; Bass and Cherrett 1994; Currie and actinomycete abundance, had higher frequencies of weed- Stuart 2001). In contrast, the functional role of fungus-

This content downloaded from 160.111.134.19 on Fri, 31 Jul 2015 11:05:40 AM All use subject to JSTOR Terms and Conditions Managing Disease in Fungus-Growing Ants 579

cultivars, and actinomycete bacteria has been likened to an arms race with respect to Escovopsis infections (Cafaro et al. 2011), with considerable emphasis on biochemistry. Our results highlight the fact that sanitary behaviors, such as weeding, grooming, and fungus planting, are under- appreciated additional weapons in the ants’ arsenal that can modify the dynamics of the race (e.g., Little et al. 2006). Further complications for understanding the dynamics of this symbiosis include factors associated with the acquisition of new fungal cultivars by farming ant populations. Relatively basal species such as Mycocepurus smithii grow at least five different cultivars (Mueller et al. 1998), while Cyphomyrmex (Green et al. 2002) and Ac- romyrmex species (Poulsen et al. 2009) may exchange fungal cultivars, although each mature attine colony appears to host only a single cultivar (Poulsen and Boomsma 2005; Mueller et al. 2010). Switching fungal cultivars increases genetic heterogeneity across colonies, which may make the Figure 4: Relationship between body size of Trachymyrmex worker symbiosis as a whole less susceptible to population-wide ants (first principal component; see text for details) and the logarithm of metapleural gland (MG) size. Least squares regression lines are sweeps of specific fungal pathogens (Hamilton 1982; shown for each species. Mueller et al. 2005). Such population structures of garden symbionts across colonies will tend to maintain genetic diversity for mutually adaptive responses of ants, fungus, planting behavior is not clear, although covering behavior and pathogens (Thompson 2005). could physically minimize the dispersion of pathogenic At present, we lack organized data on the genetic di- conidia. In addition, such behavior may inhibit germi- versity of symbiont strains cultivated by Trachymyrmex nation/growth of the pathogen by modifying microenvironmental conditions to render them unfavorable to de- velopment of the pathogen, analogous to the growth of tumors under some microenvironmental conditions but not others (Polyak et al. 2008). The latter hypothesis is consistent with the fact that some fungal cultivar strains inhibit the growth of some Escovopsis strains (fig. 3B; Ge- rardo et al. 2004, 2006), although the mechanism by which they do so is not known. An understanding of the trade-offs among different defensive strategies by attines is impeded by limited data on (1) the costs of the diverse behavioral and biochemical strategies using a common currency such as metabolism, (2) dose-dependent efficacy of different antimicrobial compounds, (3) differing degrees of virulence among strains of pathogens, and (4) evolutionary lability of behavioral and biochemical defenses. Synthesis of MG secretions accounts for more than 15% of basal metabolic energy in Acromyrmex (Poulsen et al. 2002), and main- taining actinomycete bacteria carries approximately the same cost (Poulsen et al. 2003a, 2003b), but nothing is known of the energetic costs of prophylactic and hygienic behaviors (Fernańdez-Marıń et al. 2007). Trade-offs among defensive strategies are most thoroughly studied Figure 5: Relationship between metapleural gland (MG) size (re- for plants’ antiherbivore defenses, which are shaped by the moving the effects of worker body size by using the residuals from the regressions in fig. 4) and actinomycete cover (after removing the relative diversity and abundance of specialist and generalist effect of differences between colonies; see text for details) for Tra- enemies (see references in “Introduction”). chymyrmex workers. The 95% confidence ellipses and reduced major Coevolution among fungus-growing ants, their fungal axes for the data are shown for each species.

This content downloaded from 160.111.134.19 on Fri, 31 Jul 2015 11:05:40 AM All use subject to JSTOR Terms and Conditions 580 The American Naturalist species in central Panama. If diversity of fungal symbionts lows. With respect to infections using Escovopsis from Tra- itself has a role in disease management, then species that chymyrmex zeteki nests, we eliminated two replicates from make minimal use of actinomycetes and MG grooming Trachymyrmex sp. 10, two replicates from T. zeteki,no (e.g., T. cornetzi and Trachymyrmex sp. 3) should have a replicates from Trachymyrmex sp. 3, and three replicates higher diversity of cultivar strains, while species that make from Trachymyrmex cornetzi. For infections using Escov- extensive use of actinomycytes or MG grooming (e.g., T. opsis from T. cornetzi nests, we eliminated two replicates zeteki and Trachymyrmex sp. 10) should have a limited from Trachymyrmex sp. 10, four replicates from T. zeteki, diversity of fungal symbiont strains. The genetic diversity one replicate from Trachymyrmex sp. 3, and two replicates of Escovopsis strains attacking Panamanian Trachymyrmex from T. cornetzi. For infections using Escovopsis from Tra- symbionts is also unknown, but provisional data suggest chymyrmex sp. 3, we eliminated six replicates from Tra- that there is probably no tight long-term association be- chymyrmex sp. 10, one replicate from T. zeteki, one rep- tween Escovopsis strains and the fungal cultivar, because licate from Trachymyrmex sp. 3, and two replicates from the same strains of Escovopsis can be found in both closely T. cornetzi. Finally, for infections using Escovopsis from and distantly related attines (e.g., Apterostigma [Gerardo Trachymyrmex sp. 10, we eliminated three replicates from et al. 2006] and Atta [Taerum et al. 2007]). Although the Trachymyrmex sp. 10, six replicates from T. zeteki, two large-scale co-occurrence patterns of attine ants, cultivars, replicates from Trachymyrmex sp. 3, and two replicates Escovopsis, and Pseudonocardia are increasingly well un- from T. cornetzi. derstood (Currie et al. 2006; Cafaro et al. 2011), there is much left to be done before we understand the local and Methods to Score Abundance of Actinomycetes regional dynamics of the host-mutualist-parasite interactions that govern the attine ant farming symbiosis. Over two years, we collected a sample of ants from different nests for each of the four study species, to make comparisons among species and between years. The number of colonies from which workers were sampled in 2007 Acknowledgments and 2008, respectively, are 13 and 11 colonies for Trachy- We thank S. Dreier, two anonymous reviewers, and J. Tewks- myrmex sp. 10, 12 and 11 colonies for T. cornetzi, 8 and bury for comments and suggestions to improve earlier 10 colonies Trachymyrmex sp. 3, and 15 and 14 colonies drafts. This work was supported by postdoctoral fellowships for Trachymyrmex zeteki. For 2007 collections, we exam- p from the Secretarıá Nacional de Ciencia, Tecnologıá y In- ined 50 workers per colony, except for T. cornetzi (n novacioń (SENACYT; Becas 2005–2010, Repu´ blica de Pan- 25 workers). In 2008, we examined 25, 16–50, 25, and 20– ama´; and the International Collaboration Program, COL09- 25 workers per colony for Trachymyrmex sp. 10, T. cornetzi, 038) and the Smithsonian Tropical Research Institute (STRI) Trachymyrmex sp. 3, and T. zeteki, respectively. and funding from the Centre for Social Evolution, De- partment of Biology, University of Copenhagen to H.F.-M. Additional support was provided by general research funds Literature Cited from STRI to W.T.W. E.B.G. and G.B. were sponsored by Adams, R. M. M., T. H. Jones, A. W. Jeter, H. H. De Fine Licht, T. the Adelante Internship program at STRI via W.T.W. J.J.B. R. Schultz, and D. R. Nash. 2012. A comparative study of exocrine and D.R.N. were supported by a grant from the Danish gland chemistry in Trachymyrmex and Sericomyrmex fungus-grow- National Research Foundation (DNRF57). We thank the ing ants. Biochemical Systematics and Ecology 40:91–97. Agrawal, A. A. 2007. Macroevolution of plant defense strategies. Autoridad Nacional del Ambiente for research and collect- Trends in Ecology and Evolution 22:103–109. ing permits and STRI staff for logistic support. Bailey, I. W. 1920. Some relations between ants and fungi. Ecology 1:174–189. Barke, J., R. F. Seipke, S. Gru¨schow, D. Heavens, N. Drou, M. J. Bibb, R. J. M. Goss, D. W. Yu, and M. I. Hutchings. 2010. A mixed community of actinomycetes produce multiple antibiotics for the APPENDIX fungus farming ant Acromyrmex octospinosus. BMC Biology 8:109. Bass, M., and J. M. Cherrett. 1994. The role of leaf-cutting ant work- Methods to Assess Inhibition of Escovopsis sp. ers (Hymenoptera: Formicidae) in fungus garden maintenance. Sporulation by the Fungal Cultivar Ecological Entomology 19:215–220. Biani, N. B., U. G. Mueller, and W. T. Wcislo. 2009. Cleaner mites: In vitro assays were used to assess whether fungal cultivars sanitary mutualism in the miniature ecosystem of Neotropical bee are able to inhibit sporulation of Escovopsis, from con- and nests. American Naturalist 173:841–847. heterospecific nests. In some cases, the petri dishes were Boomsma, J. J., P. Schmid-Hempel, and W. O. H. Hughes. 2005. Life overgrown by contaminants and were discarded, as fol- histories and parasite pressure across the major groups of social

This content downloaded from 160.111.134.19 on Fri, 31 Jul 2015 11:05:40 AM All use subject to JSTOR Terms and Conditions Managing Disease in Fungus-Growing Ants 581

insects. Pages 139–175 in M. D. E. Fellowes, G. J. Holloway, and establishment by Acromyrmex ants (Hymenoptera, Formicidae, At- J. Rolf, eds. Insect evolutionary ecology. CABI, Wallingford. tini). Insectes Sociaux 54:64–69. Bot, A. N. M., D. Ortius-Lechner, K. Finster, R. Maile, and J. J. Fincher, R. M., L. A. Dyer, C. D. Dodson, J. L. Richards, M. A. Tobler, Boomsma. 2002. Variable sensitivity of fungi and bacteria to com- J. Searcy, J. E. Mather, A. J. Reid, J. S. Rolig, and W. Pidcock. 2008. pounds produced by the metapleural glands of leaf-cutting ants. Inter- and intraspecific comparisons of antiherbivore defenses in Insectes Sociaux 49:363–370. three species of rainforest understory shrubs. Journal of Chemical Cafaro, M. J., and C. R. Currie. 2005. Phylogenetic analysis of mu- Ecology 34:558–574. tualistic filamentous bacteria associated with fungus-growing ants. Fine, P. V. A., Z. H. Miller, I. Mesones, S. Irazuzta, H. M. Appel, M. Canadian Journal of Microbiology 51:441–446. H. H. Stevens, I. Saä¨ksja¨rvi, J. C. Schultz, and P. D. Coley. 2006. Cafaro, M. J., M. Poulsen, A. E. F. Little, S. L. Price, N. M. Gerardo, The growth-defense trade-off and habitat specialization by plants B. Wong, A. E. Stuart, B. Larget, P. Abbot, and C. R. Currie. 2011. in Amazonian forests. Ecology 87:S150–S152. Specificity in the symbiotic association between fungus-growing Gerardo, N. M., C. R. Currie, S. L. Price, and U. G. Mueller. 2004. ants and protective Pseudonocardia bacteria. Proceedings of the Exploitation of a mutualism: specialization of fungal parasites on Royal Society B: Biological Sciences 278:1814–1822. cultivars in the attine ant symbiosis. Proceedings of the Royal Coley, P. D., and J. A. Barone. 1996. Herbivory and plant defenses Society B: Biological Sciences 271:1791–1798. in tropical forests. Annual Review of Ecology and Systematics 27: Gerardo, N. M., U. G. Mueller, and C. R. Currie. 2006. Complex 305–335. host-pathogen coevolution in the Apterostigma fungus-growing Condit, R., S. Aguilar, A. Hernandez, R. Perez, S. Lao, G. Angher, ant-microbe symbiosis. BMC Evolutionary Biology 6:88. S. P. Hubbell, and R. B. Foster. 2004. Tropical forest dynamics Green, A. M., R. M. M. Adams, and U. G. Mueller. 2002. Extensive across a rainfall gradient and the impact of an El Ninõ dry season. exchange of fungal cultivars between sympatric species of fungus- Journal of Tropical Ecology 20:51–72. growing ants. Molecular Ecology 11:191–195. Cremer, S., S. A. Armitage, and P. Schmid-Hempel. 2007. Social Haeder, A., R. Wirth, H. Herz, and D. Spiteller. 2009. Candicidin- immunity. Current Biology 17:R693–R702. producing Streptomyces support leaf-cutting ants to protect their Cremer, S., and M. Sixt. 2009. Analogies in the evolution of individual fungus garden against the pathogenic fungus Escovopsis. Proceed- and social immunity. Philosophical Transactions of the Royal So- ings of the National Academy of Sciences of the USA 106:4742– ciety B: Biological Sciences 364:129–142. 4746. Currie, C. R., U. G. Mueller, and D. Malloch. 1999a. The agricultural Hamilton, W. D. 1982. Pathogens as causes of genetic diversity in pathology of ant fungus gardens. Proceedings of the National their host populations. Pages 269–296 in R. M. Anderson and R. Academy of Sciences of the USA 96:7998–8002. M. May, eds. Population biology of infectious diseases. Springer, Currie, C. R., M. Poulsen, J. Mendenhall, J. J. Boomsma, and J. Berlin. Billen. 2006. Coevolved crypts and exocrine glands support mu- Hart, A. G., A. N. Bot, and M. J. Brown. 2002. A colony-level response tualistic bacteria in fungus-growing ants. Science 311:81–83. to disease control in a leaf-cutting ant. Naturwissenschaften 89: Currie, C. R., J. A. Scott, C. R. Summerbell, and D. Malloch. 1999b. 275–277. Fungus-growing ants use antibiotic-producing bacteria to control Heil, M. 2002. Ecological costs of induced resistance. Current Opin- garden parasites. Nature 398:701–704. ion in Plant Biology 5:345–350. Currie, C. R., and A. E. Stuart. 2001. Weeding and grooming of Hughes, W. O. H., and J. J. Boomsma. 2004. Genetic diversity and pathogens in agriculture by ants. Proceedings of the Royal Society disease resistance in leaf-cutting ant societies. Evolution 58:1251– B: Biological Sciences 268:1033–1039. 1260. Currie, C. R., B. Wong, A. E. Stuart, T. R. Schultz, S. A. Rehner, U. ———. 2006. Does genetic diversity hinder parasite evolution in G. Mueller, G. H. Sung, J. W. Spatafora, and N. A. Straus. 2003. social insect colonies? Journal of Evolutionary Biology 19:132–143. Ancient tripartite coevolution in the attine ant–microbe symbiosis. Hughes, W. O. H., R. Pagliarini, H. B. Madsen, M. B. Dijkstra, and Science 299:386–388. J. J. Boomsma. 2008. Antimicrobial defense shows an abrupt evo- Dijkstra, M. B., and J. J. Boomsma. 2003. Gnamptogenys hartmani lutionary transition in the fungus-growing ants. Evolution 62: Wheeler (Ponerinae: Ectatommini): an agro-predator of Trachy- 1252–1257. myrmex and Sericomyrmex fungus-growing ants. Naturwissen- Itino, T., and T. Itioka. 2001. Interspecific variation and ontogenetic schaften 90:568–571. change antiherbivore defense in myrmecophytic Macaranga. Eco- Fernańdez-Marıń, H., J. K. Zimmerman, D. R. Nash, J. J. Boomsma, logical Research 16:765–774. and W. T. Wcislo. 2009. Reduced biological control and enhanced Jander, R. 1976. Grooming and pollen manipulation in bees (Apoi- chemical pest management in the evolution of fungus-farming in dea): the nature and evolution of movements involving the foreleg. ants. Proceedings of the Royal Society B: Biological Sciences 276: Physiological Entomology 1:179–194. 2263–2269. Kost, C., T. Lakatos, I. Bo¨ttcher, W.-R. Arendholz, M. Redenbach, Fernańdez-Marıń, H., J. K. Zimmerman, S. A. Rehner, and W. T. and R. Wirth. 2007. Non-specific association between filamentous Wcislo. 2006. Active use of metapleural glands by ants in con- bacteria and fungus-growing ants. Naturwissenschaften 94:821– trolling fungal infection. Proceedings of the Royal Society B: Bi- 828. ological Sciences 273:1689–1695. Kozo-Polyansky, B. (1924) 2010. Symbiogenesis: a new principle of Fernańdez-Marıń, H., J. K. Zimmerman, and W. T. Wcislo. 2004. evolution. V. Fet, trans. V. Fet and L. Margulis, eds. Harvard Ecological traits and evolutionary sequence of nest establishment University Press, Cambridge, MA. Original edition, Puchina, in fungus-growing ants (Hymenoptera, Formicidae, Attini). Bio- Leningrad. logical Journal of the Linnean Society 81:39–48. Lee, K. A., L. B. Martin II, D. Hasselquist, R. E. Ricklefs, and M. ———. 2007. Fungus garden platforms improve hygiene during nest Wikelski. 2006. Contrasting adaptive immune defenses and blood

This content downloaded from 160.111.134.19 on Fri, 31 Jul 2015 11:05:40 AM All use subject to JSTOR Terms and Conditions 582 The American Naturalist

parasite prevalence in closely related Passer sparrows. Oecologia Poulsen, M., A. N. M. Bot, and J. J. Boomsma. 2003a. The effect of (Berlin) 150:383–392. metapleural gland secretion on the growth of a mutualistic bac- Leong, W., and J. R. Pawlik. 2011. Comparison of reproductive pat- terium on the cuticle of leaf-cutting ants. Naturwissenschaften 90: terns among 7 Caribbean sponge species does not reveal a resource 406–409. trade-off with chemical defenses. Journal of Experimental Marine Poulsen, M., A. N. M. Bot, C. R. Currie, M. G. Nielsen, and J. J. Biology and Ecology 401:80–84. Boomsma. 2003b. Within-colony transmission and the cost of a Little, A. E. F., and C. R. Currie. 2007. Symbiotic complexity: dis- mutualistic bacterium in the leaf-cutting ant Acromyrmex octo- covery of a fifth symbiont in the attine ant-microbe symbiosis. spinosus. Functional Ecology 17:260–269. Biology Letters 3:501–504. Poulsen, M., A. N. M. Bot, M. G. Nielsen, and J. J. Boomsma. 2002. ———. 2008. Black yeast symbionts compromise the efficiency of Experimental evidence for the cost and hygienic significance of antibiotic defenses in fungus-growing ants. Ecology 89:1216–1222. the antibiotic metapleural gland secretions in leaf-cutting ants. Little, A. E. F., T. Murakami, U. G. Mueller, and C. R. Currie. 2003. Behavioral Ecology and Sociobiology 52:151–157. Construction, maintenance, and microbial ecology of fungus-grow- Poulsen, M., H. Fernańdez-Marıń, C. R. Currie, and J. J. Boomsma. ing ant infrabuccal pellet piles. Naturwissenschaften 90:558–562. 2009. Ephemeral windows of opportunity for horizontal trans- ———. 2006. Defending against parasites: fungus-growing ants com- mission of fungal symbionts in leaf-cutting ants. Evolution 63: bine specialized behaviours and microbial symbionts to protect 2235–2247. their fungus garden. Biology Letters 2:12–16. Reber, A. R., G. Castella, P. Christe, and M. Chapuisat. 2008. Ex- Mattila, H. R., and T. D. Seeley. 2007. Genetic diversity in honey bee perimentally increased group diversity improves disease resistance colonies enhances productivity and fitness. Science 317:362–364. in an ant species. Ecology Letters 11:682–689. Mattoso, T. C., D. D. O. Moreira, and R. I. Samuels. 2012. Symbiotic Sachs, J. L., and E. L. Simms. 2006. Pathways to mutualism break- bacteria on the cuticle of the leaf-cutting ant Acromyrmex subter- down. Trends in Ecology and Evolution 21:585–592, doi:10.1016 raneus subterraneus protect workers from attack by entomopath- /j.tree.2006.06.018. ogenic fungi. Biology Letters 8:461–464. Schmid-Hempel, P. 1998. Parasites in social insects. Princeton Uni- Mueller, U. G. 2002. Ant versus fungus versus mutualism: ant-cultivar versity Press, Princeton, NJ. conflict and the deconstruction of the attine ant-fungus symbiosis. Schoenian, I., M. Spiteller, M. Ghaste, R. Wirth, H. Herz, and D. American Naturalist 160(suppl.):S67–S98. Spiteller. 2011. Chemical basis of the synergism and antagonism Mueller, U. G., D. Dash, C. Rabeling, and A. Rodrigues. 2008. Co- in microbial communities in the nests of leaf-cutting ants. Pro- evolution between attine ants and actinomycete bacteria: a re- ceedings of the National Academy of Sciences of the USA 108: evaluation. Evolution 62:2894–2912. 1955–1960. Mueller, U. G., N. M. Gerardo, D. K. Aanen, D. L. Six, and T. R. Schultz, T. R., and S. G. Brady. 2008. Major evolutionary transitions Schultz. 2005. The evolution of agriculture in insects. Annual Re- in ant agriculture. Proceedings of the National Academy of Sci- view of Ecology, Evolution, and Systematics 36:563–595. ences of the USA 105:5435–5440. Mueller, U. G., H. Ishak, J. C. Lee, R. Sen, and R. R. Gutell. 2010. Sen, R., H. D. Ishak, D. Estrada, S. E. Dowd, E. Hong, and U. G. Placement of attine ant-associated Pseudonocardia in a global Pseu- Mueller. 2009. Generalized antifungal activity and 454-screening donocardia phylogeny (Pseudonocardiaceae, Actinomycetales): a of Pseudonocardia and Amycolatopsis bacteria in nests of fungus- test of two symbiont-association models. Antonie van Leeuwen- growing ants. Proceedings of the National Academy of Sciences hoek 98:195–212. of the USA 106:17805–17810. Mueller, U. G., S. A. Rehner, and T. D. Schultz. 1998. The evolution Stamp, N. 2003. Out of the quagmire of plant defense hypotheses. of agriculture in ants. Science 281:2034–2038. Quarterly Review of Biology 78:23–55. Oh, D. C., M. Poulsen, C. R. Currie, and J. Clardy. 2009. Denti- Taerum, S. J., M. J. Cafaro, A. E. Little, T. R. Schultz, and C. R. gerumycin: a bacterial mediator of an ant-fungus symbiosis. Na- Currie. 2007. Low host-pathogen specificity in the leaf-cutting ant– ture Chemical Biology 5:391–393. microbe symbiosis. Proceedings of the Royal Society B: Biological Oi, D. H., and R. M. Pereira. 1993. Ant behavior and microbial path- Sciences 274:1971–1978. ogens (Hymenoptera: Formicidae). Florida Entomologist 76:63–74. Tarpy, D. R., and T. D. Seeley. 2006. Lower disease infections in Pe´rez-Orte´ga, B., H. Fernańdez-Marıń, M. S. Loiaćono, P. Galgani, honeybee (Apis mellifera) colonies headed by polyandrous vs mon- and W. T. Wcislo. 2010. Biology notes on a fungus-growing ant, androus queens. Naturwissenschaften 93:195–199. Trachymyrmex cf. zeteki (Hymenoptera, Formicidae, Attini) at- Thompson, J. N. 2005. The geographic mosaic of coevolution. Uni- tacked by a diverse community of parasitoid wasps (Hymenoptera, versity of Chicago Press, Chicago. Diapriidae). Insectes Sociaux 57:317–322. Weber, N. A. 1972. Gardening ants, the attines. American Philo- Polyak, K., I. Haviv, and I. G. Campbell. 2008. Co-evolution of tumor sophical Society, Philadelphia. cells and their microenvironment. Trends in Genetics 25:30–38. Poulsen, M., and J. J. Boomsma. 2005. Mutualistic fungi control crop Associate Editor: Joshua Tewksbury diversity in fungus-growing ants. Science 307:741–744. Editor: Mark A. McPeek

This content downloaded from 160.111.134.19 on Fri, 31 Jul 2015 11:05:40 AM All use subject to JSTOR Terms and Conditions